Hi Peggy,
1. All anatomical volumes used in the last update of longitudinal tracula are in the same space as the *.long.* freesurfer volumes, i.e., the within-subject base template.
2. There are no changes to the bedpostx step.
2a. The brain mask should be the same across all steps of trac-all, i.e., either use a mask based on the anatomical segmentation throughout, or use a mask based on the low-b diffusion volumes throughout. Whether you want to use the anatomical or diffusion brain mask for another analysis with FSL is up to you.
3. You can't skip the tensor fit step. Even if it weren't used it in the -prior step, the -path step would need it because it calculates average FA and other such measures for each tract that it reconstructs. In any case, skipping the tensor fit step would not save you much time.
Hope this helps, a.y
On Mon, 25 Aug 2014, Peggy Skelly wrote:
I have a few questions about the May 2014 update to Tracula:
- Between the 2013-12-04 trac-all version 1.53 and 2014-05-26 trac-all
version 1.56, the longitudinal stream now uses the aparc+aseg.mgz from the *.long.* directories, and all output is saved there, too. Are the <subj_tp>.long.<subj_base>/dlabel/anat and anatorig transformed into the base space like other images from the longitudinal recon-all, or are they in the original space of the timepoint T1 image?
- Are there any changes to the bedpostX step, other than using a different
mask? I've already run bedpostx in the Dec 2013 version (with lowb_brain_mask). Can the results be used with the new version by moving the <subj>/dmri.bedpostX directory to <subj_tp>.long.base/dmri.bedpostX?
2a. (Maybe pertains to both versions) The link to .../dmri/nodif_brain_mask.nii.gz seems to be made during the -prep stage using .../dlabel/diff/aparc+aseg_mask.bbr.nii.gz or lowb_brain_mask.nii.gz depending on usemaskanat's value in the dmrirc file. The bedpost stage seems to use the target file as its mask no matter what usemaskanat is in the dmrirc file used when calling trac-all -bedp. (I'm trying to use aparc+aseg_mask for -prep and -path, but lowb_brain_mask for -bedp in anticipation of using the bedpostx output for other seed-base tracking in FSL.)
- May 2014 version: trac-all -prep -notensor fails when it gets to
computing the priors because dmri_train includes the --fa flag and the dtifit_FA.nii.gz is not present. Running trac-all -prior without a computed FA also fails the same way. I could compute the tensor, but thought I would save some time. I didn't realize computing the priors would use it. Does it improve the accuracy?
Thanks, Peggy